Sales of shares of artificial intelligence technology suppliers on Tuesday were the clearest sign that sentiment in this decade-long trend is changing. Concerns are growing that rising interest rates could undermine the spending frenzy, which surpasses the scale of America's historical railway boom.
Over four years since OpenAI launched ChatGPT, capital has poured into semiconductor companies, data center owners, cable manufacturers, and all others benefiting from the unprecedented race to build computing infrastructure.
Cracks in these investments, which began to appear in June, are now spreading across various markets: last week, a data center initial public offering collapsed in Australia, and credit investors have become more selective regarding AI-related debt.
In Asia, sales of shares of the South Korean chip manufacturer continued for the fifth consecutive day on Tuesday and accelerated sharply. The Kospi index fell by more than 6%, and in Tokyo, memory maker Kioxia led the losses, dropping 13% due to AI.
These declines are the most significant since the collapse of similar stocks at the end of July. The Kospi index is on its longest streak of declines in over two years and has fallen by a quarter of its July highs, while the Star Market tech market in Shanghai lost more than a third of its July peaks.
Damien Boyd, a portfolio strategist at Wilson Asset Management in Sydney, noted: 'The most important development was not only the rise in bond yields but also bond volatility... and this is starting to reflect on asset prices across all sectors.' He added: 'This is a very uncomfortable situation in many aspects... Perhaps AI capital expenditures will just slow down a bit, and some players will be victims of this.'
Massive Sell-offs
The Philadelphia Semiconductor Index, an American benchmark, fell by 2.7% on Monday. Even stronger selling led to drops of 10% and 8% respectively for Samsung Electronics and SK Hynix, listed in Seoul. Samsung's fall occurred a few days after the company forecasted record quarterly operating profit of $80 billion amid a global memory chip shortage.
On Monday, the research division of BlackRock, the world's largest asset manager, downgraded its outlook on US stocks to neutral, citing 'potential volatility around AI infrastructure buildout,' and became optimistic about short-term US Treasury bonds.
The International Monetary Fund (IMF) warned in its latest Global Financial Stability Report on Tuesday that 'the consequences of the AI correction could be global.' This came a day after its Managing Director, Kristalina Georgieva, stated that AI investments support global growth by about 3%. The IMF acknowledged AI's contribution to growth but noted growing evidence that investment potential and economic benefits may be less than expected.
Caution is also spreading to primary markets. Last week, Firmus, an Australian data center operator supported by Nvidia, postponed a $5 billion IPO due to waning investor interest, highlighting how quickly risk has become the main focus for investors channeling capital into this sector.
The sale of Tencent Holdings' dollar and 'dim sum' bonds—debt obligations denominated in Chinese yuan and issued in Hong Kong—was the latest deal under scrutiny in debt markets on Tuesday. Its forecast for 10-year dollar notes was about 1.15 percentage points above the yield on US Treasury bonds, which stand at 5.2%.
Last month, Japanese tech investor SoftBank sold the largest high-yield bond deal in history, offering up to 9.75% on a $4.5 billion deal with a seven-and-a-half-year maturity.
Shifting sentiment dominated several major financial conferences in Singapore, particularly on Monday when Broadcom's CEO stated that capital is the main constraint on growth.
Some do not believe the AI trend is completely over—Meta's shares rose 3.4% on Monday after releasing a new AI model. However, many view semiconductors as a warning sign.
Mark Velan, Head of Investments at Lucerne Asset Management in Singapore, suggested that the 'harsh session' in Korea says more about positioning and the cost of capital than any fundamental deterioration in AI demand. He added: 'I suspect we are entering the next phase of the AI cycle, where markets are increasingly focusing on monetization and return on invested capital—I wouldn't call it the end of the AI trend. But Korea could be an early warning of a broader shock.'